The objectives of this section are especially clearcut. They include familiarity with the definition of the Laplace transform .e{f{t)} = F(s) that is given in Equation (1) in the textbook, the direct application of this definition to calculate Laplace transforms of simple functions (as in Examples 1-3), and the use of known transforms (those listed in Figure 7.1.2) to find Laplace transforms and inverse transforms (as in Examples 4-6). Perhaps students need to be told explicitly to memorize the transforms that are listed in the short table that appears in Figure 7.1.2.
1.
(u=-st, du=-sdt)
= f-oo[_1 ] ue" du = _!_[(u -1)eu]-'" = _1
S2 S2 0 S2
o
2. We substitute u = -st in the tabulated integral
f u'e" du = eU (u2 - 2u + 2) + C (or, alternatively, integrate by parts) and get
3.
.e{e31+1} = r e-.I'le31+1 dt = e r e-(.I'-3)1 dt = .E: s-3
4.
With a = -s and b = 1 the tabulated integral
f au b d au [acosbu + bSinbU] C
e cos u u = e 2 2 +
a «b
yields
.e{ } r -.1'1 d. [e-.I'I (-scost + sint)]'"
cos t = e cos t t = 2 =
S + 1 1=0
S
S2 + 1 .
Section 7.1
379
5. .e{sinht} = t.e{el _e-t} = t r e-sl(et -e-I)dt = t r (e-(S-I)1 _e-(s+l)t)dt
= ~ [ s ~ 1 - S: 1] = S2 ~ 1
7.
8.
9.
10.
11.
12.
13.
14.
15.
1 1 e-s
= ---+-
S S2 S2
.e{Ji +3t} = r(~:22) +3.~ = ~2 +~
S S 2s s
.e{t-2e31} = l_ __ 2_
S2 s-3
.e{t3/2+e-lOt} = r(5/2) +_1_ = 3J; +_1_
S512 s + 10 4S5/2 S + 10
.e{l +cosh5t} = !..+ 2 S
S S -25
380
Chapter 7
16.
17.
18.
19.
20.
{ . } 2 s s+2
£. sm2t + cos2t = --+-- = --
S2 + 4 i + 4 S2 + 4
£.{cos22t} = ..!..£.{I+cos4t} = ..!..(..!..+ 2 s )
= lim( 2e-SI 2 (-SCoskt+ksinkt)]- 2 eO 2 (-s.l+k.O) = s
1->00 S + k s + k S2 + k2 •
382
Chapter 7
36. Evidently the function f(t) = sine e12) is of exponential order because it is bounded; we can simply take c = 0 and M= 1 in Eq. (23) of this section in the text. However,
its derivative J' (t) = 2t el2 cos( e12) is not bounded by any exponential function eCI, because et2 I eel = e12-CI ~ 00 as t ~ 00.
37. J(t) = 1 - ua(t) = 1 - u(t - a) so
For the graph of f, note that f(a) = 1- u(a) = 1-1 = O.
38. J(t) = u(t -a) - u(t - b),so
+as =hs
.e{f(t)} = .e{ua(t)}-.e{ub(t)} = _e e _ = S-I(e-as _e-hS).
S S
For the graph off, notethat f(a)=u(0)-u(a-b)=1-0=1 because a c b, but J(b)=u(b-a)-u(O)=I-I=O.
42. Let's refer to (n -1, n] as an odd interval if the integer n is odd, and even interval if n is even. Then our function h(t) has the value a on odd intervals, the value b on even intervals. Now the unit step function J(t) of Problem 40 has the value 1 on odd intervals, the value 0 on even intervals. Hence the function (a - b)J(t) has the value (a - b) on odd intervals, the value 0 on even intervals. Finally, the function
(a - b) f (t) + b has the value (a - b) + b = a on odd intervals, the value b on even intervals, and hence (a -b)f(t) + b = h(t). Therefore
a-b b a+be-.I'
L{h(t)} = L{(a-b)f(t)}+L{b} = _. +- = --
s(1 + e .1) s s(1 + e -'\')
SECTION 7.2
TRANSFORMATION OF INITIAL VALUE PROBLEMS
The focus of this section is on the use of transforms of derivatives (Theorem 1) to solve initial value problems (as in Examples 1 and 2). Transforms of integrals (Theorem 2) appear less frequently in practice, and the extension of Theorem 1 at the end of Section 7.2 may be considered entirely optional (except perhaps for electrical engineering students).
In Problems 1-10 we give first the transformed differential equation, then the transform Xes) of the solution, and finally the inverse transform x(t) of Xes).
25. With fit) = cos kt and F(s) = sl(i + lC), Theorem 1 in this section yields
£{-ksinkt} = £{f'(t)} = sF(s)-1 = s- s -1 = _ e ,
i+e i+e
so division by -k yields £{sinkt} = kl(i+lC).
Section 7.2
389
26. With j(t) = sinh kt and F(s) = k/(i - f(-), Theorem 1 yields
.e{f'(t)} = .e{k cosh kt} = ks/(i - k-) = sF(s) ,
so it follows upon division by k that .e{ cosh kt} = s / (i - f(-).
27.
(a)
With fit) = feat and /'(t) = nf-Ieat + at'e", Theorem 1 yields
so
and hence
(b)
n =1:
.e{ at}_ 1 .e{at}_ 1 1_
te - -- e - -_._- - --7"
s-a s-a s-a (s-a)2
1
n = 2: .e{t2 eat} = _2_.e{teat} = _2_. 1 = 2!
s-a s-a (s-a)2 (s-a)3
n--3' . .e{ 3 at} 3.e{ 2 at} 3 2! 3!
t e = -- t e = --. = ---
s-a s-a (s-a)3 (s-a)4
And so forth.
28. Problems 28 and 30 are the trigonometric and hyperbolic versions of essentially the same computation. For Problem 30 we let fit) = t cosh kt, so f(0) = O. Then
f'(t) = cosh kt + kt sinh kt f"(t) = 2k sinh kt + f(-t cosh kt,
and thus f'(O) = 1, so Formula (5) in this section yields
.e{2k sinh kt + f(-t cosh kt} = i .e{t cosh kt} - 1,
k 12 2
2k· 2 2 + K F(s) = s F(s) - 1.
s -k
We readily solve this last equation for
29. Let fit) = t sinh kt, so f(0) = O. Then
390 Chapter 7
f'(t) = sinh kt + kt cosh kt f"(t) = 2k cosh kt + it sinh kt,
and thus f'(O) = 0, so Formula (5) in this section yields
.e{2k cosh kt + it sinh kt} = i .e{sinh kt}, Zk- 2 s 2 + iF(s) = iF(s).
s -k
We readily solve this last equation for
30. See Problem 28.
31. Using the known transform of sin kt and the Problem 28 transform of t cos kt, we obtain
32. If j(t) = u(t - a), then the only jump in j(t) is ji = 1 at ti = a. Since j(0) = 0 and f'(t) = 0, Formula (21) in this section yields
o = s F(s) - 0 - easel).
Hence .e{u(t - a)} = F(s) = S-I e",
33. f(t) = Ua(t)-ub(t) = u(t-a)-u(t-b),so the result of Problem 32 gives
e-as e-bs e-as _ «':
.e{f(t)} = .e{u(t-a)}-.e{u(t-b)} = --- = --
s s s
34. The square wave function of Figure 7.2.9 has a sequence {tn} of jumps with tn = n and jn = 2( -lr for n = 1,2,3, .... Hence Formula (21) yields
00
o = s F(s) - 1 - Le-"s ·2(-1)".
11=1
It follows that
Section 7.2
391
ao
S F(s) = 1 + 2 ~)-l)"e-II.1'
11=1
-1 +2(1_e-s+e-2s_e-3s+ ... )
-1 + 21(1 + e -S)
= (1 - e-S)/(l + e-S)
= (es/2 _ e-s/2)/(es/2 + e-s/2)
s F(s) = tanh(s 12),
because 2 cosh(s/2) = es/2 + e-s12 and 2 sinh(s/2) = es12 - e-s12.
35. Let's write get) for the on-off function of this problem to distinguish it from the square wave function of Problem 34. Then comparison of Figures 7.2.9 and 7.2.10 makes it clear that get) = +(1 + f(t)), so (using the result of Problem 34) we obtain
1 1 1 1 s 1 ( e.l'/2 - e-.I'/2 «:" J
G(s) = -+-F(s) = -+-tanh- = - 1+. . '-.-
2s 2 2s 2s 2 2s esl2 + e-,,12 e-·,/2
= _1 (1 + 1- e -.I' J = _1. 2 = 1 .
2s l+e-s 2s l+e-.I' s(l+e-'\')
36. If get) is the triangular wave function of Figure 7.2.11 and fit) is the square wave function of Problem 34, then g'(t) = f(t). Hence Theorem 1 and the result of Problem 34 yield
£ {g'(t)} = s £{g(t)} - g(O),
F(s) = s G(s), (because g(O) = 0)
£{g(t)} = S-l F(s) = s-2tanh(s/2).
37. We observe that f(O) = 0 and that the sawtooth function has jump -1 at each of the points til = n = 1, 2,3,···. Also, f'(t) == 1 wherever the derivative is defined. Hence Eq. (21) in this section gives
1 ao ao 1
- = sF ( s ) + Ie-lis = sF ( s ) - 1 + Ie -11.1' = s F ( s ) - 1 + -liS '
S 11=1 11=0 1- e
ao
using the geometric series Ix" = 1/(1- x) with x = «: Solution for F(s) gives
11=0
392
Chapter 7
SECTION 7.3
TRANSLATION AND PARTIAL FRACTIONS
This section is devoted to the computational nuts and bolts of the staple technique for the inversion of Laplace transforms - partial fraction decompositions. If time does not permit going further in this chapter, Sections 7.1-7.3 provide a self-contained introduction to Laplace transforms that suffices for the most common elementary applications.
When we multiply both sides by the quadratic factor S2 + 2s + 2 and collect coefficients, we get the linear equations
-2B-D+3 = 0
-2A-2B-C = 0
-2A-B+1 = 0
-A = 0
which we solve for A = 0, B = 1, C = -2, D = 1. Thus
F( ) = 1 -2s + 1 = 1 _ 2. s + 1 3. 1
s 2 + 2 2 2+ 2'
(s + 1) + 1 [( s + 1)2 + 1] (s + 1) + 1 [ (s + 1)2 + 1] [( s + 1)2 + 1]
We now use the inverse Laplace transforms given in Eq. (16) and (17) of Section 7.3- supplying the factor e-I corresponding to the translation s ~ s + 1 - and get
When we multiply each side by the quadratic factor (squared) 5 we get the identity
2s3 - i = (As + B)(4i - 4s + 5) + Cs + D.
When we substitute the root s = 1/2 + i of the quadratic into this identity, we find that C = -3/2 and D = - 5/4. When we first differentiate each side of the identity and then substitute the root, we find that A = 1/2 and B = 114. Writing
4i - 4s + 5 = 4[(s - 1I2i + 1],
it follows that
1 (s-t)+1 1 3(s-t)+4
F(s) = -. 2 -_. 2'
8 (s - t) + 1 32 [( s - t)2 + 1 ]
Finally the results
.e-1 {2s/(i + Ii} = t sin t
.e-1 {2/(i + Ii} = sin t - t cos t
ofEqs. (16) and (17) in Section 7.3, together with the translation theorem, yield
I( ) 1/2 [I ( inr] 3 1. 4 1 ( . )]
t = e _. cost+smt --·-tsmt--·- smt-tcost
8 32 2 32 2
= _1 eI/2[(8+4t)cost+(4-3t)sintJ.
64
23.
and i ± 2as + 2a2 = (s ± ai + a2, so it follows that
DERIVATIVES, INTEGRALS, AND PRODUCTS OF TRANSFORMS
This section completes the presentation of the standard "operational properties" of Laplace transforms, the most important one here being the convolution property £{f*g} = £{f}.£{g}, where the convolution f' g is defined by
f * g(t) = f f(x)g(t - x) dx.
Here we use x rather than r as the variable of integration; compare with Eq. (3) in Section 7.4 of the textbook.
Section 7.3
401
1. With J(t) = t and g(t) = 1 we calculate
[1 I
t*l = {x.ldx = _x2
2 x=o
1 2 = -t.
2
2. With J(t) = t and get) = eat we calculate
f-( ) ( d) at
at Uu U e u
= e _ - a e - -;; = 7 [ue du
(with u = -ax)
eat
= -[(u-1)e"J-
a' -
(integral formula #46 inside back cover)
• m
= ~[( -ax _1)e-·axJx=t . = ~[( -at _1)e-at + 1J
a2 x=o a2
t * eat = -\- ( eat - at - 1).
a
3. Tocompute (sint)*(sint) = lsinxsin(t-x)dx, we first apply the identity sin A sin B = [cos(A - B) - cos(A + B)]/2. This gives
(sint)*(sint) = f sinxsin(t-x)dx 1 rt
= 21 [cos(2x-t)-cost]dx
1 [1 . ]x=t
= - -sm(2x-t)-xcost
2 2 x~
(sint)*(sint) = _!_(sint-tcost).
2
4. To compute t2 * cost = l' x2 cos(t - x)dx, we first substitute
cos(t - x) = cos t cos x + sin t sin x,
and then use the integral formulas
fx2 cosx dx = x2 sinx + 2xcosx - 2sinx+ C
fx2 sinx dx = - x2 cosx + 2xsinx + 2cosx+ C.
402
Chapter 7
from #40 and #41 inside the back cover of the textbook. This gives
28. An empirical approach works best with this one. We can construct transforms with powers of (i + 1) in their denominators by differentiating the transforms of sin t and cos t. Thus,
From the first and last of these formulas it follows readily that
Alternatively, one could work out the repeated convolution
In Problems 31-42, we first write and transform the appropriate differential equation. Then we solve for the transform of the solution, and finally inverse transform to find the desired solution.
x(t) = get) + 2I( -1)" Un;r (t)g(t - mr) = 2sin2 t +4I( -1)"un;r(t)sin2 t.
n=1
11=1
Hence
x(t) = {2sin2t if 2mZ'st«2n+l)7£, -2sin2 t if (2n -1)7£ s t < 2n7£.
Consequently the complete solution
x(t) = 21sintlsint
is periodic, so the transient solution is zero. The graph of x(t):
x (t)
2
t
r:
420
Chapter 7
42. x" + 2x' + lOx = j(t), x(O) = x'(O) = 0
As in the solution of Example 8 we find first that
so
lOco 1 O( -1)" e-mrs
Xes) = +2L .
S(S2 +2s+10) 11=1 S(S2 +2s+10)
If
get) = £1{ [ 10, J} = 1-!e-I(3cos3t+sin3t),
s (s+l) +9 3
then it follows that
co
x(t) = g(t)+2L(-1)"umr(t)g(t-mr).
The graph of x(t):
x (t)
2
-2
SECTION 7.6
11=1
2
IMPULSES AND DELTA FUNCTIONS
4
6
t
Among the several ways of introducing delta functions, we consider the physical approach of the first two pages of this section to be the most tangible one for elementary students. Whatever the
Section 7.5
421
approach, however, the practical consequences are the same - as described in the discussion associated with equations (11 )-(19) in the text. That is, in order to solve a differential equation of the form
a x"(t) + b x'(t) + c x(t) = J(t)
where fit) involves delta functions, we transform the equation using the operational principle .e{ oa(t)} = e =. then solve for Xes), and finally invert as usual to find the formal solution x(t). Then we show the graph of the inverse transform x(t).
1. iX(s) + 4X(s) = 1 X
1 1
Xes) = -2-
s +4
1 . 2 1
x(t) = -sm t 2
2
t
If
1
2
-1 2. iX(s) + 4X(s) = 1 + e-7lS